Efferent ducts (ED) are a series of small and highly convoluted tubes that connect the testis with the epididymis. It is generally accepted that ED originate from the mesonephric tubules. During mouse embryonic development, the mesonephric tubules regress and, in females, they disappear. In males, this regression is not complete and the remaining tubules, which are connected to the cranial side of the Wolffian duct (Staack et al.,2003), form the ED under the influence of testosterone produced by the developing testis (Sainio et al.,1997).
These tubes have been frequently considered as a simple “way out” from the gonad. However, since the first detailed description by Becker in 1856, it is known that the epithelium of ED is the only one along the male reproductive tract which is lined by true ciliated cells (Ilio and Hess,1994), suggesting an active role of ED in sperm cell transit. In the last decade, several genes have been found to be important in the physiology and morphogenesis of mouse ED. Disruption of genes participating in water transport as Na+/H+ transporter and aquaporin-1 produce phenotypes where the accumulation of testicular fluid is due to a decrease of liquid absorption by the ED epithelium (Zhou et al.,2001). It has been shown that most of the water, electrolytes, and proteins of the testicular fluid is reabsorbed by the ED epithelium (Clulow et al.,1994). Estrogen receptor alpha, which is expressed in the ED, was shown to be a master gene in the control of this process (Hess et al.,1997).
We have recently reported that LGR4, a G protein-coupled receptor closely related to LH, FSH, and TSH receptors, is important for the postnatal development of the male reproductive tract (Mendive et al.,2006). In lgr4 knock-out (LGR4KO) mouse, the epididymis fails to elongate leading to a poorly convoluted and highly dilated tract. We have also found that the transit of sperm cells is physically blocked somewhere in the ED, as suggested by the massive accumulation of spermatozoa in the rete testis. In histological preparations of ED, we have observed that sperm cells accumulate exclusively at the “proximal” or “testis” side of the duct network. It has been reported that blind-ended ED (i.e., ED that do not connect to the epididymis) often explain sperm accumulation or “spermiostasis” within the rete testis of several species (Ilio and Hess,1994; Hess et al.,2000). This observation raised the possibility that a similar developmental defect might occur in LGR4KO.
In the early characterization of this phenotype, we performed dissection under a binocular microscope of the ED of five wild-type (WT) and five LGR4KO adult mice as well as examined histological sections of the ED of numerous WT, heterozygote, and LGR4KO animals (minimum 10 sections spread through the duct tree in at least 10 adult animals in each condition; Mendive et al.,2006). We did not observe macroscopical or histological evidence of blind ending tubes (Ilio and Hess,1994; Hess et al.,2000). However, we found small segments of duct stenosis or atresia could not be entirely ruled out by these techniques. Consequently, we decided to use a computer-assisted method to construct from thin serial histological sections a full 3D model of the LGR4KO ED tree in order to improve our understanding of the nature of the obstacle to the sperm flow.
MATERIAL AND METHODS
The 3D models described in this article were constructed from 2D serial histological images of one WT and one KO organ. Importantly, the main characteristics observed in the KO organ used to build the model (sperm accumulation in rete testis and proximal ED, dilation of proximal ducts, presence of syncytial-like bodies in the luminal space), were observed reproducibly in each LGR4KO mouse analyzed (Mendive et al.,2006). This guarantees that the animal used to construct the model is indeed representative of the LGR4KO phenotype.
WT and LGR4KO male reproductive tracts (testis-ED-epididymis; on CD1 genetic background) were dissected, embedded in paraffin, and sectioned transversally (8 μm thickness), from the cranial pole of the testis up to the rete region. This way of sectioning guaranteed that the whole ED tree and its connections with the testis and epididymis were included in the series. All sections were stained with haematoxylin-eosin and photographed at a 25× magnification with a CCD camera (SPOT RT, Diagnostic Instruments, http://www.diaginc.com) mounted on a Zeiss microscope (Axioplan2, Carl Zeiss Inc., http://www.zeiss.com). All animal experiments were approved by the Institutional Animal Use and Care Committee and conducted in accordance with the International Guiding Principles for Biomedical Research Involving Animals as promulgated by the Society for the Study of Reproduction.
Assembling the picture stack
The pictures were manually aligned one by one using Adobe Photoshop CS (Adobe Systems, http://www.adobe.com). The picture of the first section was used as background (Fig. 1A) onto which a picture of the adjacent section was superimposed, as a second layer (Fig. 1B). This “layer 2” was made semi-transparent and aligned using the “transform-rotate and transform-distort” commands to correct the slight distortions due to handling of the tissue during paraffin embedding and sectioning (Fig. 1C). This allowed a reliable overlaying of the consecutive pictures. When the alignment was optimal, “layer 2” was made opaque again (Fig. 1D) and the same procedure was applied to the next picture, and so on for the whole set of 360 pictures. Each aligned layer was saved afterwards in an individual jpeg file numbered according to its position along the z axis. This series of jpeg images corresponds to the image stack needed to perform 3D reconstruction (Fig. 1E, right panel).
3D-doctor (Able Software Corp., http://www.ablesw.com/3d-doctor) was used for the rendering of the 3D model. The stack of 360 images was loaded in the program and the true voxel size and the units (μm) were entered in the calibration menu. The image pixel size (x = y = 3.7 μm), defined both by the resolution of the camera (1520 × 1080 pixels) and the microscopic field surface, was measured by taking the picture of a scale bar graduated at intervals of 0.01 mm. The slide thickness “z” was set on 8 μm on the microtome while cutting the sample.
The first step of model construction was to outline every ED cross-section present on each tissue section. One cross-section, selected randomly at the testicular end of the efferent duct, was outlined using “free-hand drawing” and the tube was followed up and down around each bend along the stack, drawing its outline in the same color toward its epididymal end (Fig. 1E, left panel). As 3D-doctor allowed adding slices to an existing stack, it was possible to start drawing from a restricted number of pictures, add the next slices progressively, and check while drawing this first tube the global quality of the alignment for any drifting occurring in the stack due to human alignment errors. Transition from one tube section to the next was unambiguous and the 3D reconstruction of this first tube showed a smooth regular appearance of all bends, thus validating the alignment of the stack. Each of the remaining tubes was then outlined from its exit point from the tunica albuginea to its junction with the next tube with a different color, each color defining a separate object in the 3D reconstruction software (Fig. 1E, left panel).
Following this process, we have found that four ED exit the gonad, both in WT and mutant animals. The collector tube that results from the junction of the four ED (represented in yellow) was drawn until its junction with the epididymis.
Once all tubes were drawn up to the beginning of the epididymis, the tube boundaries were smoothened using the “boundary process-smooth boundary” command, reducing the distance between nodes to 10, to lighten the demand on the computer processor when generating the surface model. A small square was duplicated from the first to the 50th picture in the z axis to provide a 400-μm long visual scale reference within the 3D model visualization window.
The available surface-rendering modes were tested; “complex rendering” gave the best results and was therefore used in all subsequent steps using the recommended parameters (x = 3, y = 3, z = 1). The model surface was smoothened three times using the “tools-smooth surface” command to reduce the “steps” generated by this mode of reconstruction and to give a more natural-looking appearance to the tubes (Fig. 1F–G).
The same arbitrary color code was assigned in the WT and KO models to the tubes identically placed in the ED tree. The LGR4KO model was imported into the WT workspace with the calibration values set in the same units for both models to allow visual comparison of their relative sizes. The addition of a scale-bar in each model ensured that no unwanted resizing had been performed by the software upon importation of the LGR4KO model (Fig. 2C, online Supporting Information Data 1).
Morphometric parameters from the 3D model
The 3D reconstruction software makes available the total surface and volume of each individual tube. Using these values, we deduced the mean radius (R) and length (L) applying the formulae: R = 2 × volume/surface; L = surface2/4πvolume. These formulae assume a simple cylindrical shape for all the tubes with a constant radius which is not the case particularly in LGR4KO where the testicular end of the tubes is clearly dilated. To obtain values which stick closer to the true length of the tubes, we built up partial models of the straight and convoluted regions.
Reconstruction of efferent ducts connections in WT and LGR4KO mice
The initial question, motivated by the massive sperm cell accumulation in the testis of LGR4KO, was: do ED connect to each other and flow into the epididymis, or are they blind-ended tubes? The 3D reconstruction allowed individualizing each tube (Fig. 2A), and tracking their way from the testis towards the epididymis. We found that four ducts exit the testis in both KO and WT animals. At the testis or “proximal” side, ED are poorly convoluted and run parallel to each other. At about 4 mm from the rete testis, they become highly convoluted and they anastomose (Fig. 2A–C, online Supporting Information Data 1). In our model, the first junction occurs between the tan and orange tube. A second junction occurs between orange and pink tube. Then, the junction of the fourth tube (blue) occurs (Fig. 2B). The collector tube that results from the junction of the four ED (represented in yellow) flows normally into the epididymis. The same anastomosing pattern is followed in WT and LGR4KO animals.
Morphometric parameters of LGR4KO and WT efferent ducts obtained from the 3D reconstruction
The 3D models allowed computation of a series of morphometric parameters of the ED which are not accessible by other methods (Fig. 2D, Supporting Information Table 1). The 3D reconstruction software makes available the total surface and volume of each individual tube. Using these values, we deduced a mean radius (R) and the length (L) of the tubes. The calculated mean radius R was 76% larger in LGR4KO than in WT (39.64 vs. 22.49 μm). This is an expected observation if we consider the sperm cell accumulation present in the mutant organ. In contrast, the overall length of the LGR4KO ED is reduced to 50% of the WT (38.23 vs. 81.90 mm).
We have previously suggested that ED of LGR4KO are less convoluted compared with the WT, based on the reduction in the number of tube cross-sections observed in histological preparations. Because the main known function of this organ is to concentrate sperm cells by absorbing water from the testicular fluid, we proposed that the low convolution of the ducts, and the accompanying reduction of the epithelium surface, might impair this process contributing in swelling of the testis and ED (Mendive et al.,2006). Murine ED display three anatomically different regions along their way from the testis to the epididymis: (1) the proximal segment, where tubules run individually parallel to each other, and are not convoluted; (2) a central segment, known as “the cone,” where tubules become highly convoluted and anastomose; and (3) the distal segment, where all the tubules have converged into a single one before connecting to the epididymis. Our 3D model clearly shows that KO ED are much less convoluted than WT (Fig. 2A,C). To quantify this observation, we built up partial models in order to calculate the length and surface of the proximal (straight) and cone (convoluted) regions separately (Fig. 2A, white arrows). The data show that the length and surface of the cone region in LGR4KO ED is one half of the corresponding in the WT (WT: 69.53 versus KO: 31.29 mm, and WT: 9.18 versus KO: 4.62 mm2. Fig. 2D and Supporting Information Table 1).
An additional clear difference between the WT and LGR4KO 3D models is the variation in the diameter of the ducts at their proximal and distal sides. While in the WT animal the mean radius of proximal and distal ducts are quite similar (28.165 and 21.167 μm, respectively), in the KO, the mean radius of the proximal ducts is much bigger than that of the distal ones (51.01 and 22.84 μm, respectively) (Fig. 2D lower panel).
The 3D reconstruction of efferent ducts helps to explain spermiostasis in LGR4KO
We have previously reported the presence of syncytial-like bodies in the luminal space of ED of LGR4KO mice, associated with sperm cell accumulation in rete testis (Mendive et al.,2006). Although these bodies seemed to completely occlude the tubular lumen, the mere examination of the 2D images was not sufficient to establish if sperm cells can pass through. To answer this question, we represented the spermatozoa accumulation and the syncytial-like bodies (Fig. 3E) in the context of the 3D model (Fig. 3A–D, online Supporting Information Data 2). In three out of the four LGR4KO ED (tan, orange, and blue tubes, Fig. 3A, B, and C, respectively), sperm cells accumulation abruptly stops where the syncytial-like bodies appear, and no trace of sperm was found further down at the epididymal side. In the forth duct (pink), sperm accumulates along the whole tube (Fig. 3D). Interestingly, the syncytial-like bodies were only found in the same three tubes in which sperm stops flowing. These findings show that the presence of the syncytial-like bodies in the ED lumen directly correlates with the blockade of the sperm cells flow, confirming our previous data obtained by trypan blue injection (Mendive et al.,2006). Additionally, we observed a 32-μm long stenosis in the distal part of the blue tube with focal complete disappearance of its luminal space (Fig. 3D,F–I). Whether this stenosis reflects a developmental abnormality or a secondary process remains unknown.
The ED anatomy has already been extensively described centuries ago (for review, Ilio and Hess,1994). ED arises from the rete testis which sits centrally in the dog, bull, goat, and boar testis but lies at the testis margin in mouse, rat, hamster, and man. After an initial testicular segment, the ED pierces the tunica albuginea and appears in the epididymal ligament as lightly convoluted and rather thick tubes. We, indeed, observed this in our 3D model. Interestingly, this model also allowed us to estimate the length and mean diameter of this straight ED segment and of the following highly convoluted section.
The ED number varies between two and 33 depending on the species. In mouse, three to five ED may be observed, while in man six to 15 can be found (Ilio and Hess,1994). In both our WT and LGR4KO, we observed four ED which indeed agrees with what could be expected for the mouse.
The terminal region of the ED, where they connect to the epididymis, presents two main designs in mammals. On the one hand, in rat, mouse, and some guinea pigs the ED form a funnel, the coni vasculosi, where they are first highly tortuous then become moderately coiled and thinner as they approach the head of the epididymis. In the coni, the ED anastomose to each other to produce a single tube that then abruptly turns into the initial segment of the epididymis. On the other hand, in man and most guinea pigs, the ED form multiple parallel coils separately entering the head of the epididymis. In large mammals, the ED organization is comparable with that observed in man, but they may join the epididymis in pairs or in groups after forming connecting tubes (for review, Ilio and Hess,1994). In our 3D model, the four ED present in both WT and LGR4KO exhibit the classically described “mouse type” connecting pattern. These data show unambiguously that LGR4KO ED connect normally and fully to each other and reach the epididymis as in the WT organ, ruling out the hypothesis that they might be blind-ended.
During the characterization of male infertility in LGR4KO mice (Mendive et al.,2006), we were confronted with the inherent limitations one faces when trying to visualize a 3D structure from 2D images. Those limitations precluded the unambiguous explanation of the observed phenotype.
The usefulness of the reconstructions of a 3D structure from 2D images has been recognized for more than a century. These early attempts were limited to the contours of the studied organ until Blechschmidt altered the procedure to allow reconstruction of different structures in one model (Arnold and Kleiner,2004). Only recently, did modern computer techniques give true accessibility and depth to 3D visualization with the possibility of electronic dissection (Sperber et al.,1996), which allows the removal of single structures from embryos for study or embryology teaching (Sperber,2003). Medical imaging techniques like CTscan, MRI, and now, 3D echography have vastly contributed to the growing popularity of 3D modeling for the study of embryonic and adult structures in normal and pathological conditions, in humans as well as in animal models.
The 3D model described in this article efficiently helped us to answer most of the initial questions that motivated its construction and constitutes, to our knowledge, the first reconstruction of the spatial structure of the complete mouse ED (Saitoh et al.,1990).
On the basis of the 2D histological images, we have proposed that LGR4KO ED are less convoluted than WT. The morphometric values obtained from the 3D model confirmed this previous observation, but also provided the quantitative data showing that the surface of the convoluted region of KO ED is 50 % of the WT. This reduction in surface of LGR4KO ED, clearly evidenced by the 3D model, fits well with the fact that epithelial cells of ED and epididymis proliferate less in LGR4KO (Mendive et al.,2006), and supports the idea that Lgr4 plays a role in the tube elongation process of the male reproductive tract.
Previous experiments also suggested that LGR4KO male infertility was partially a consequence of an obstructive azoospermia produced by a blockade of sperm flow, somewhere in the ED tree (Mendive et al.,2006). By allowing an integration of histological data (the content in the lumen of each tube) with the reconstructed 3D anatomy of the organ, the model provided a more precise location of the blockade, as well as information about the structures involved. In three of the four ducts, syncytial-like bodies were found to constitute an obstacle, in association with upstream accumulation of sperm, while in the fourth duct, where no syncytial-like body was found, sperm cells are observed up to the end of the tube.
In normal mice, however, sperm transit is very active in this region of the reproductive tract and only very few spermatozoa are normally found in the ED lumen. It has been proposed that ED smooth muscle layer peristaltism and not the beat of the cilia of the ED ciliate cells is the major contributor to fluid flow from the testis to the epididymis. Fluid viscosity—and thus sperm concentration and water reabsorption—is an important factor as well in this model (Winet,1980). The fact that large amounts of spermatozoa were found in the lumen of LGR4KO ED, even in the tube where no visible obstacle was found at histology, suggests that LGR4 might also play a role in regulating the flow of sperm from the testis to the epididymis. This regulation could occur at different levels and through multiple mechanisms since previous in situ hybridization and LacZ staining data have shown lgr4 to be expressed not only in ED cells where it could affect fluid reabsorption and cilliary beat but possibly also in the ED very thin outer smooth muscle layer where it could affect ED peristaltism (Mendive et al.,2006 and unpublished data). An alternative (but not exclusive) hypothesis would be that the reduction of sperm flow in the ED could be secondary to (or worsened by) the developmental anomaly present in the epididymis (Mendive et al.,2006). According to this hypothesis, the syncytial-like structures could be an early consequence of diminished sperm flow, rather than the cause. This would provide an explanation to the observation that one duct in our model displays no anatomical blockade, despite showing sperm accumulation up to its end.
A recent study has reported that male knockout mice carrying a deletion of the He6 gene (Human Epididymal protein 6 or Gpr64) are infertile due to a dysregulation of fluid absorption within the ED (Davies et al.,2004). Interestingly, large amounts of spermatozoa accumulate in the ED of these animals as in LGR4KO. A careful examination of the ED histology shown in this article (Fig. 4B in Davies et al.,2004) also reveals the presence of the syncytial-like bodies. This observation strongly supports that these particular histological structures might originate as a secondary reaction to the sperm transit defect within the ED, aggravating the sperm stasis by creating a physical blockade. In agreement with this hypothesis, we have previously shown that these syncytial-like bodies are composed, at least partially, of cells of immune origin, as suggested by the expression of the pan-leukocyte antigen CD45 (Mendive et al.,2006).
In a recent article, Hoshii et al. (2007) have confirmed our previous data showing that LGR4 plays a key role in the development of the mouse epididymis (Mendive et al.,2006). Even if the main aspects of the phenotype are conserved between the two mutant models, they did not observe the sperm transit defect discussed here (and the resulting sperm accumulation in rete testis and ED systematically present in our animals). This difference can be explained considering our previous observation that the LGR4 deficiency phenotype is dependent on the genetic background, also pointed out by Hoshii et al. (2007) when they show a clear variation in postnatal mortality when the mutant allele is expressed in different mouse strains.
In WT rat, tight junctions between adjacent ED nonciliated cells have been found to be segmented or incomplete (Suzuki and Nagano,1978). The presence of such leaky junctional complexes as well as observations of intraepithelial lymphocytes in the normal rat ED (Dym and Romrell,1975) suggested that the permeability barrier of this epithelium in the blood-to-lumen direction is already weak in wild type animals. One additional and interesting aspect of the phenotype reported by Hoshii et al. (2007) is the distortion and multilamination of the basal membrane, characterized by an abnormal laminin accumulation which leaks into the epididymal lumen. This phenomenon merits further investigation but it might be an indication of an especially weakened epithelium lining in the reproductive tract in LGR4KO. We have previously proposed that the syncytial-like aggregates made of immunitary cells, present in the ED lumen, are a consequence of a defective epithelium which cannot establish its barrier function failing to isolate sperm cells from the immune system (Mendive et al.,2006). The finding that the basal membrane in the LGR4 mutant mouse is abnormal, reinforce our explanation about the origin of the syncytial-like aggregates and guide future investigations to elucidate the mechanisms underlying the morphogenetic defects in LGR4 deficiency.
Obstructive azoospermia is a frequent cause of infertility in the human population accounting for 30–40% of all cases (Ezeh,2000). The obstacle to the sperm flow may be located in the testis (presence of sperm agglutinins), at the epididymis level (absence or distended epididymis) or beyond (absence of vas deferens). Localization of the obstacle is important for the selection of the appropriate treatment of the infertility: percutaneous or microsurgical epididymal sperm aspiration when epididymal spermatozoa are available, testicular sperm aspiration, in cases of absence or fibrosis of the epididymis (Ezeh,2000). However, current imaging techniques have not permitted to estimate the prevalence of ED agenesis or blockade in obstructive spermiostasis in men. Although it has been widely accepted that genetic abnormalities are involved in many cases of defective spermatogenesis, the role of genetic diseases in case of obstructive azoospermia is probably underestimated (Ezeh,2000). This is highlighted by the fact that up to 60% of men with congenital absence of vas deferens have mutations in the CFTR gene (Chillon et al.,1995). Although the phenotype of LGR4KO mice includes additional characteristics beyond male infertility (intrauterine growth retardation, body size reduction) (Mazerbourg et al.,2004; Mendive et al.,2006) and notwithstanding the fact that contrary to what is observed in mouse, rat, and some guinea pigs, in human and larger mammals ED enter the epididymis at multiple sites and do not form a single distal duct (Ilio and Hess,1994), the present study opens the possibility that LGR4 might be responsible for certain cases of human male infertility. Possible candidates should be searched for amongst patients in whom sperm cells accumulate in the rete testis and ED, but not in the epididymis.
LGR4 belongs to Leucine-rich G-protein-coupled Receptors (LGR) subfamily of membrane receptors characterized by a large ectodomain containing leucine-rich repeat motifs, in addition to the canonical heptahelical serpentine domain typical of GPCRs (Hsu et al.,1998). In the last years, new members of this subfamily have been identified in various species along the animal kingdom, from Anthopleura elegantissima to human (Costagliola et al.,2005; Kudo et al.,2000; Loh et al.,2001). In the particular case of the new mammalian LGRs, it is becoming evident that they are not only structurally related to the founder members of the family (the FSH and LH/hCG receptors), but they also play a role in the same physiological processes. It was recently shown that LGR7 and LGR8 are the natural receptors of relaxin (Hsu et al.,2002), a hormone with key functions in mammalian reproduction. In addition, mutations in Lgr8 produce infertility associated with cryptorchidism in mice (Overbeek et al.,2001). The male infertility observed in LGR4KO mice places Lgr4 as a new member of the family involved in mammalian reproduction and again points out the importance of the LGR subfamily of GPCR in reproductive biology. The question regarding the nature of the endogenous agonists of the orphan LGRs, still remains open. The study of mice strains harboring LGRs null alleles might be useful tools to answer this question.
Computer-assisted 3D reconstruction from 2D histological sections is a powerful tool to study the morphology of organs or embryonic structures which are too small or hidden, to be macroscopically appreciated. The finding of a short 32-μm long stenosis in our LGR4KO model does not seem relevant to explain the LGR4KO phenotype. However, it is an interesting feature to discuss with regards to the use of 3D modeling to study truly small anatomic structures. This stenosis would have almost certainly been overlooked in 2D but could have been lost to 3D too if the interval between slices had been higher. As it is we had first tried to reconstruct the 3D models based on microscopic sections with 24 μm interval, only to find that the manual alignment of the sections could not be done unambiguously in the highly convoluted distal portion of the tubes with this z distance. This points to the fact that the interval between slices must be carefully chosen with regards to the size of the anomaly one wishes to be able to examine.
In the anomaly of the ED reported in LGR4KO mice, 3D reconstruction helped to understand the “structure-function” relationships of the diseased organ, which had not been possible from mere 2D histology. In addition, it allowed calculating morphometric parameters such as total volume, surface, radius, and length of the ED that are not accessible by other methods.
Combined with immunodetection or RNA in-situ hybridisation, 3D anatomical reconstruction from serial histological sections constitutes a powerful tool to study morphogenetic processes and organ function in normal and pathological conditions. Reconstruction of several KO and WT ED systems would have been mandatory if one had wanted to obtain truly absolute morphometric parameters with confidence interval accounting for individual and technical variability as well as statistical analyze of the difference between WT and LGR4KO. Yet given the important amount of time needed to generate histology-based high resolution 3D models, and the fact that careful 2D observation of the testis, ED, and epididymis of the LGR4KO mouse used for the 3D model showed lesions similar to the phenotype we described previously (Mendive et al.,2006), we felt safe drawing conclusions from only one KO and WT models. In the global study of the LGR4KO phenotype, it might be interesting to reconstruct the ED at various times during embryonic development to determine precisely when the anomaly in ED length/surface starts to show and how this feature correlates with the intraluminal blockade, inflammatory infiltrate, and spermiostatsis.
The authors thank Mr Ted Wu from Able Software Corporation for the easy and efficient communication all along the development of our 3D model. This fruitful interaction resulted in a number of useful upgrades of the software which fitted progressively better to our needs. PV and M-AL are Research Associate and Research Fellow of the FNRS, respectively.